Location: Office of Associate Administrator
Project Number: 0500-00093-001-011-S
Project Type: Non-Assistance Cooperative Agreement
Start Date: Sep 30, 2024
End Date: Sep 29, 2026
Objective:
The use of unoccupied aerial systems (UAS) to rapidly collect imagery has the potential to greatly accelerate agricultural research. However, successful application of this technology requires timely processing of raw imagery from UAS into analysis-ready data. The purpose of this Non-Assistance Cooperative Agreement (NACA) is to support ongoing cooperation and collaboration between the United States Department of Agriculture (USDA) Agricultural Research Service (ARS) and North Dakota State University (NDSU) for the development of UAS image processing software that is compatible with, and available on, ARS’s high-performance computing (HPC) infrastructure that is a part of ARS’s Scientific Computing Initiative (SCINet). SCINet is an effort by ARS to enhance the USDA’s research capacity by providing scientists with access to HPC clusters, cloud computing, a high-capacity data storage environment, advanced networking for data transfer, and training in scientific computing.
Approach:
ARS has deep scientific expertise across a wide spectrum of agricultural research domains, many of which can benefit from the rapid collection of imagery by UAS. ARS also has the computing infrastructure to support high-throughput processing of UAS imagery, including multiple HPC clusters. The Cooperator has major research programs in agriculture and natural resources and extensive expertise in leveraging HPC and modern computing and data analytics for scientific research. This NACA will support the collaborative extension of UAS imagery processing software, AgSkySight, that is in-development by the Cooperator’s Agricultural Data Analytics group (AgDA), to run on SCINet’s HPC infrastructure. AgSkySight provides an efficient and user-friendly workflow for UAS data ingestion, image stitching, and analysis, particularly for agricultural research plots. The software consists of four components, all of which will be implemented on SCINet infrastructure: a web frontend, backend API (application programming interface), an open-source processing engine, and an imagery database. This software will equip ARS researchers with the ability to pre-process the UAS imagery they are collecting or, in many cases, have already collected but are unable to process in reasonable timeframes with local computational resources. Although the processing engine used by AgSkySight can already be run on SCINet clusters, the additional three components in AgSkySight will significantly reduce the barrier of entry for ARS researchers to migrate their UAS image processing workflows to the HPC environment. This will provide ARS researchers with user-friendly access to the computational resources required to process their UAS imagery in reasonable timeframes.
